Corner plots

Produce corner plots for a single sources’ parameters.

Load catalog and select individual source

import matplotlib.pyplot as plt
import numpy as np

from lisacattools.catalog import GWCatalog
from lisacattools.catalog import GWCatalogs
from lisacattools.catalog import GWCatalogType

# load catalog
catPath = "../../tutorial/data/ucb"
catalogs = GWCatalogs.create(GWCatalogType.UCB, catPath, "cat15728640_v2.h5")
final_catalog = catalogs.get_last_catalog()
detections_attr = final_catalog.get_attr_detections()
detections = final_catalog.get_detections(detections_attr)

# get index of source with maximum SNR
sourceId = detections.index.values[
    np.argmin(np.abs(np.array(detections["SNR"]) - detections["SNR"].max()))
]
detections.loc[[sourceId], ["SNR", "Frequency"]]

	SNR	Frequency
name
LDC0092117281	969.357	0.009212

Corner plot of select source parameters using corner module

import corner

# read in the chain samples for this source
samples = final_catalog.get_source_samples(sourceId)

# list of subset of paramters that are particularly interesting
parameters = ["Frequency", "Frequency Derivative", "Amplitude", "Inclination"]

# corner plot of source.
fig = corner.corner(samples[parameters])

Can also be done with ChainConsumer for prettier plots

from chainconsumer import ChainConsumer

# get dataframe into numpy array (this shouldn't be necessary)
df = samples[parameters].values

# rescale columns
df[:, 0] = df[:, 0] * 1000  # f in mHz
df[:, 1] = df[:, 1] / 1e-15  # df/dt
df[:, 2] = df[:, 2] / 1e-22  # A
df[:, 3] = df[:, 3] * 180.0 / np.pi  # inc in deg
parameter_symbols = [
    r"$f\ {\rm [mHz]}$",
    r"$\dot{f}\ [s^{-2}]\times 10^{-16}$",
    r"$\mathcal{A} \times 10^{-23}$",
    r"$\iota\ {\rm [deg]}$",
]

# add source chain to ChainConsumer object
c = ChainConsumer().add_chain(df, parameters=parameter_symbols, cloud=True)

# plot!
fig = c.plotter.plot(figsize=1.5)
plt.show()